How "Pooled Testing" Could Drastically Improve Virus Testing Capabilities

Listeners to the Ben Greenfield Podcast, this is Matt Dawson and Mike Mallon from the Wild Health Podcast. We've been talking to Ben recently about an idea that we've been very excited about working on for the last couple of weeks. Ben is excited about as well, and he asked us to record something for you. So I think this is probably. I don't. I don't remember being as excited about anything for quite a while. There's a couple things that we know right now with the pandemic and everything that's going on, we have to open back up at some point, and we have to do it safely. Mike, you saw the meatpacking plan in Waterloo, Iowa, that recently had a thousand people test positive for Covid.

Yeah. That's insane.

Out of about 3,000 employees. And this just can't happen. Like, there are essential workers right now who are still working. A lot of us are getting ready to go back to work, and we can't have this happening. And it's going to happen if we have asymptomatic spreaders. And we think the solution to that is testing.

You can't just test everybody. You can't expect to test the entire population. Think about how much that would cost. Think about the resources that we don't even have for the numbers of tests that we need for sick patients right now. I mean, that's just unrealistic.

Right? So in the current regime and how we were doing it, that would have been unrealistic, and that's why we're so excited about this idea. So I want to give you an example of what pulled testing. And this is not a new concept, something that's been around for a long time, but it's time to apply it to the novel coronavirus. So if you say you have a thousand employees that you want to screen and make sure they don't have Covid, either you're going to bring them back to work, or you want to do routine screening to make sure you have an asymptomatic spreader come in. So you have a thousand employees. And let's say that the prevalence right now, if you're screening asymptomatic people, say it's one in 100. It's actually probably more like one in a thousand, but let's say it's one in 100. So if you take those thousand individuals, you swab them, but then you group them into pools of 10, and you take 10 swabs, put all those in the same test tube, and then you run that one RT PCR the CDC gold standard test. If you have 10 people that are positive, so say 1% of 1000, you're going to get 10 pools that test positive. So 10 pools of 10 is 100 people. So you've rolled out 900 of the 1000 just by running those initial 100 tests. So then the 100 you haven't rolled out, you test those individually and you've ran 200 tests to screen 1,000 people. So you get to the point where you can do this and rule out large numbers of people efficiently and cost effectively. And if our example, if what we believe to be true, which is asymptomatic people with that are, you're just screening, are one in a thousand, then you're going to get one tube that pops positive. It'll be 110 tests. So the prevalence determines how many tests you do. But you can see how that would work, for example, with a 300 employer business, 30 pools of 10, you run 30 tests. If you have three positives, one in 100, which is a high number, then you rerun those 30 individuals by themselves and you've ran 60 total tests instead of 300. You can even.

The first time you talked to me about this, my initial thought was this, this just doesn't work. There's no way that you could just, you could put all those swabs into one, one vial and actually run it. But it turns out that's not true. Right?

Well, you don't put them in one vial. So this is actually something that happens in the lab. So in the lab, if you take all of these and you go ahead and fill a 96 well plate and you do the extraction, you, you can then do the pulling. You take 10 of those out and you run one test on that. So it's actually done in the lab. You could certainly do 10 swabs and put them in one tube. We explain it that way just so people kind of get a better understanding. Because most people can't picture what the lab is like and what the extraction equipment is like and what the PCR is like. But that's the general concept.

This has been proven in history. Is that right?

Yeah. And you could use an even bigger example. So when you say proven in history, we've. In the 1940s, actually there was a mathematician that came up with this idea because we were screening men for syphilis going into the military in World War II. And it's a low prevalence thing. So why not pull 10 together and run them and rule out 10 men at a time. And then once a pool tests positive, you can test all 10. We found that we could definitely decrease that. And you can see how this scenario could work even for a 9,000 plant facility. 9,000 people, you can do 300 pools of 30. There's some complicated math around the right pool size. But if you can do pools of 30, if you have one in a thousand people, then if 9,000 are screened and nine are positive, then you're going to be left with pools of 30 that test positive and 270 individuals. Now, let's say it's a bad situation again. You have 1% prevalence or 90 individuals that are positive. If 90 are positive, then 90 pulls a 30 test positive after one round. That's 2,700 individuals. You've still screened out 6,300. And then you do the second test. At that point, you have 90 individuals positive out of 2,700. So that's a higher prevalence. So you have to do a smaller pool size. You would have to do pools of 10. So 270 pools of 10. You test those again, you get at a worst case scenario, 90 pools that test positive. You could get lucky and have 2, 3, 4, 5 people in one pool that are testing positive. Then you have even fewer to retest and you've ruled out more. But worst case scenario, you have 90 pools of 10 or 900. So in the third round of testing, you simply test all 900 of those. So what you've done, if you had 1% prevalence, is you ran 1147 tests to rule out 9,000 individuals. And you could do that with the right equipment in less than 24 hours.

So that was a ton of math, Matt. That was really complicated. I had trouble listening without my eyes glazing over. And I like math. But what you're basically saying is by pulling these tests together, you can take what would be 9,000 individual tests and you can instead run pools. And when an entire pool is negative, that means all of those people are negative and you extract them from the rest of the possible positives. And then you just rerun test on people who in the pools that were positive. So you're reducing, in some cases with a low prevalence, the amount of tests that you're having to perform by like 80 to 90%. Is that right?

That's right. And if you can't follow the math, that's fine. We actually have a full document that explains this with all the math written out and multiple examples. If you go to wildhealth.com that document that we wrote that has the references to everything that we're talking about is going to be there. So because there are a few studies on this as well. And so if you have the, the 9,000 person example, Mike, then you can see how. And just to be clear about this, the time for this isn't now, but in say six months. Let's say we want to have a concert or a sporting event. If that's the case and we feel comfortable with it, the math works. Let's just use the example of you want to buy a $100 ticket to a concert and they're going to take.

A $270 ticket to Pearl Jam. That was on April 17 that I didn't get to go to.

So let's say you want to go to Pearl jam and it's $270, nor you're probably okay paying $300 and go ahead and paying $30 for the test. So what you could do in that circumstance, especially if you realize you're going to be one of 9,000 instead of one of 30,000 at the event. So let's say you go on. Yeah, so you pay $300, you get an email saying you have a ticket. Now you cannot get in unless you have your electronic ticket and you have your negative test. So we're part, we've partnered with a company that tracks this with an app. So what happens then? The person's notified, they go into the app and it says you have to get tested the day before, not the week before, because we don't want time to elapse where you could be exposed. And the testing is at the site, at the arena. So you set up 50 drive throughs at the arena that you've staffed and the person says where then they can schedule what time they're going to show up and to which booth and it gives them directions. They show up, they get swabbed. All of these go back to the lab. The lab then runs the test. And in 24 hours with pool testing, you can rule these people out. And the first pool of testing, we said 6,300 people were ruled out. At a minimum, those 6,300 get a notification on their app. They're negative by the time of the concert the next day, then everyone will have had a positive or negative on their app. And to get in, you show your ticket and you show the app negative. So again, not time for that. Right now we're talking about getting essential workers back to work. But you can see how you really can decrease the risk. And we can Start to get back to a little bit of normalcy with pool testing, because that's not a lot of resources. $30 test.

One thing though, Matt, the first thing that came to mind when you started talking about this stuff was, you know, is it possible that we're just basically diluting out the samples and we're going to get false, more false negatives because of dilution?

That was one of our first questions when we started looking at this. I called our lab director, Dr. Bright. We started talking about. He had that question and it seemed like the answer was no, this could work. We found a paper out of Israel where they're doing this, a paper out of Stanford where they validated the method. And then actually just a few days ago, after we started talking about the Lancet published a paper showing that you could pull up to 30. And it works just fine with PCR sampling. You amplify the material and so you have to have a certain amount of rna, but you can dilute some. And we talk about what's called cycle threshold to make sure the sensitivity isn't decreasing. And for pools of up to 30, the cycle threshold works just fine. It doesn't decrease the sensitivity at all. And there also is some really nice math around what is the right pool size. And it's really extremely rare that the right pool size would be above 30 anyway. So we're always going to say 30 or less in practicality. And if you're interested in that math, it's in the paper. For the optimal pool size, it's the square root of the total that you want to test divided by the number of infected. So 300 example I gave. You want to test 300. If you have a 1% prevalence and 3 people are infected, 300 divided by 3 is 100. The square root is 10. So the optimal pool size is 10 for that example. Now if you said you had a 1 in 1000 prevalence and we'll use the thousand person example, 1000 divided by 1 is actually, let's use the 900 person example because it's easier math. So 900 people, one person has it. 1 in 900. 900 divided by 1 is 900. The square root of that is 30. So the optimum pool size would be 30. So for a 0.1% prevalence, 1 in a thousand, the pool size is about 30. So the pool size is usually going to be between 30 on a high level, low prevalence down to 5 or 6. When you have slightly higher prevalences, it's.

About as low as it can go. If you still want to save money. And yeah, I've been, I've been staring at this for, for quite a while. The prevalence tends to work all the way from, you know, 0.01% all the way up to about 25% pretty consistently with multiple different sizes of groups. So that's, I mean, 25% is, you're pushing it at that point, but there's still, you know, an incremental savings in number of tests perform.

Yeah, but you're talking about math and not real world. So the.01% prevalence, the right pull size would be 100, but you can't do 100, so you're always going to do 30. So 30 is your maximum pull. The 25% prevalence, you really don't gain any efficiency more than just doing pulling two tests or so at a time. So it's really only useful in the population, the asymptomatic folks that we're talking about screening. So 5% or less. Even when you get to 4 and 5%, you don't gain a whole lot with pulling. So.

It'S really asymptomatic people. This is not a hospital type situation. And I think we also need to clarify what test we're talking about. We've mentioned it briefly, but this is not the serology, the serology test. Lots of people have talked about serology tests and antibody testing being the key to getting us back to work, but that's not the case right now. It could be useful in some places, like I'm talking to a group in New York, they want to do this pull testing routinely and for them. If they want to test 1,000 people regularly, they could run antibody tests first. Identify the 25 people, 25% of people, or 250 that are IgG positive. Instead of testing that group every day or every week, they could have a lighter schedule because we assume they're going to be immune. So maybe they test them once a month just to make sure and everybody else they do once a week. But if you're in Kentucky or other places with that's had low exposure to run the serology test first to find the 1 in 100 people or 5 in 100 people, it's probably not cost effective. You're probably better off just doing the screening. And there is the IGM positive and some people think by testing for IGM we're ruling people out, but you're not. What you want to catch is the asymptomatic spreaders and IgM usually is not positive to have symptoms so it's not the goal. If you identify your goal, I want to catch asymptomatic spreaders at an event or work, then you got to use the right test, and that's the CDC RT PCR test. And it's got to be ran in a high complexity lab. But there is excess capacity out there. Some people may be thinking, well, if we don't have enough for hospitals, how can we do this? We have capacity in our lab alone. We could screen tens of thousands of more people. And there are other labs out there that could be doing this as well. The time wasn't for this a month ago, but it is right now. There's capacity to do this and get us back to work.

Now, Matt, you were saying that the savings is incremental for the 5% prevalence. And that's probably because the number of tests is near half of what you would normally be performing. But if actually perform that and repeat the test once, so you actually perform the test twice on every person. You're spending about the same amount of money doing this, about the same number of tests, yet now you're getting more sensitivity. You're improving the sensitivity of the test and basically reducing false negatives. So there's definitely an improvement or the opportunity for improvement there. And that's one thing that I'm really excited about about pull testing. Because, you know, you mentioned these, these events like concerts or maybe going back to work in an office where it' as critical that we catch every single person. But if you assume that there's a 1% false negative rate, then that's not acceptable. If you're talking about like a surgical center or you're talking about a mosh pit of a concert where people are having really close interaction. So how do we use this pool of testing and actually take it to the next level and improve the sensitivity of the test itself?

Yeah, well, if you can run more tests for the same amount of money, then yes, you could just do repeat testing to improve the sensitivity and have less false negatives. So it could be ran in a situation where you have 5 to 10% prevalence and you could use it that way. You wouldn't save money, but you would increase sensitivity. And you also, I mean, it's a great question you bring up about false negatives. Because we should talk about risk. And I hate talking about risk, though, without defining it and putting numbers on it. I think it's a waste of time because the fact of the matter is nothing we ever do is going to have zero risk when we get up and we get in our car and drive to work. We're at risk of dying in a car accident. So let's talk about the numbers and the sensitivity of this and risk. So let's use the big example of the 9,000 person concert. What is the real risk there? So the sensitivity of the RT PCR test in the literature is approaching 99% with a LOD, a limit of detection of 15 particles. If you have 25, 30 particles, it's a really good test. Now some people have said that in the real world the sensitivity may be more like 80%. So we really don't know. So let's talk about both scenarios. So we said we're screening asymptomatic people and we assume one in a thousand is going to have this. So if we have 99% sensitivity, then you have to test 100,000 people to miss one. So what if the sensitivity is down to 80%? Well then you have to screen 5,000 people to miss one. And so in our 9,000 person example, we probably would miss two people if it's 80% sensitivity and we probably miss zero people if it's 99% sensitivity. So let's think about that second, is that acceptable to miss two people at a 9,000 person event? Well, the R naught is high of this normally, but we're not in normal times. Everyone understands we're scared, we're wearing masks, we're social distancing, we're washing hands for doing all the right things. In the Pearl Jam example you gave, if they're having this in a 30,000 seat facility or a 50,000 seat facility and they're spreading people out like they're doing on planes now by separating seats and everyone's wearing masks and we're doing a good job. The R naught is probably more like 1 or 1.1 somewhere around there. More like influenza or lower. So if that's the case, if two people get in, then that means they may spread it to two others. Technically the R naught is like total. How many are going to spread it to? So we're talking worst case, how many they're going to spread it to in a matter of a few hours. Probably no one, but worst case it's two. You don't get an explosion from having a large number of people there without masks. That's the problem. You've got this number down to a couple people and you're taking precautions. You don't get an explosion. So let's define this risk anymore. Let's say it's got a, we have a 1% mortality. We don't know if it's going to end up being a 0.5 the infection fatality rate or one and a half. But let's say 1%, I think that's probably a little high in the end for what we're going to find out eventually. So let's say two people catch this and they have a 1% mortality rate. So if we have a 99% sensitivity in catching this and we assume one in a thousand people are positive and we assume we get the R naught down to kind of 1.1 by social distancing, I think all of those are fair assumptions. But there are assumptions, then that means 10 million people you'd have to screen before you end up getting to a death. If the 80% sensitivity number, we're talking about 500,000 people for an event. If we just exclude elderly or high risk individuals, then you're certainly in the millions of people that you would be screening and treating this way before you get to any point of talking about a mortality now that is lower risk than car accidents with millions of people driving somewhere, terrorism or other random events. So we cannot eliminate risk, but we also cannot operate on fear. And testing with this method dramatically mitigates the risk and in my mind gets it down to an acceptable point where we can do the essential jobs we need and start to get back to a level of normalcy.

You made a critical point there, Matt, that I just want to highlight because I'm not sure that everybody will have caught it. And that's because it took me a few minutes to realize that you were making it. And that's that you're assuming the prevalence in this equation of a sensitivity of 99%. So you know, basically if everyone had the disease and you had a sensitivity of 90, then you would have one false negative, right? But you're assuming that not everyone has the disease. You're assuming that the prevalence is true here and that there's one in a thousand people who actually have the disease. And that's why the sensitivity can perform so well. And that's a very critical point when you're talking about test characteristics. If you know the prevalence of disease or you have some assemblance of the prevalence of disease, then you can more applicably apply the characteristics of the test and actually talk about true numbers. And that's what is so critical here. So the thing though that drives me crazy is that how is it no one else is talking about this? I mean, this has got to be like one of the biggest potential opportunities for public health right now that I can even think of.

Well, you made a couple of great points I want to go back to. So you can definitely do the math and the calculations if 1 in 100 or 1 in 50 or anything else. So you can calculate this out if you just have to take some assumptions. And our assumptions are getting better and better. We're doing this right now for individuals. So we're starting to see how many asymptomatic people and this can't be symptomatic people. There are multiple caveats to this, but we can calculate the risk. We don't just have to be scared or panicked and we don't have to be helpless. We can know what's going on and we're getting better and better at defining those numbers. So why is no one talking about this? Well, we talked earlier. This is not a new concept. Been doing it since World War II. Israel has been doing this. Some Stanford University, as I mentioned, has published a paper. The Lancet has now published a paper. And if you. Again, we got all of this written down, the math, all the references in an article for you if you want to read through this. And then actually today the New York Times published an article. So people are talking about it now.

I think the reason why we hadn't been talking about it is because it wasn't the time. We were trying to get tests to symptomatic people and our positive rates have been like 20% in some places. So it's not the time for pull testing. You can't do this in the emergency department or the ICU where people are coming in with symptoms. Now the focus, not the focus is shifting, but we're starting to think about some other things as well. Not just containing this and treating people who have it, but also how are we going to get back to work. And now is the time to actually talk about this. And we also just didn't have capacity before, but now multiple labs have kind of been spun up. There are excess tests, there's lots of drive throughs in different states that have been created that are not getting filled. They're time slots. And like I mentioned, our lab, our one lab can screen tens of thousands of people. So if other labs join us, then we can really solve this problem. So we really want to start this conversation, start talking about this. And if you're a lab out there and you want to do this, we're happy to share our protocols and help you. You can Email us. We set up an email testingildhealth.com if you're a company or group that wants to send you back to workers, back to work safely or discuss this or have an event in the future, you can also reach out to us@wildhealth.com we're actually going to be releasing a whole course on how to do this. We've started creating reopening protocols for several businesses that have asked us to do testing and help them. So we're going to anonymize those, share those protocols with you. This is a time I think for all of us to share everything that we have and get the information out. So we're going to be posting all this@wildhealth.com I would say. Very important point. Follow the guidance of your local officials. A lot of things that we're talking about right now are forward looking that we're not ready. You may be ready in some places and not ready in others. So whether or not this is the time for you and your business or an event where you are. I don't know when you're listening to this and where you're listening to this, but when it is time, we're happy to help and we're happy to help anyone else who wants to be able to help in those regions. Basically we're an open book. We're happy to share. Just let us know how we can help. We've got to get open back up and we've got to do it safely.